我國各選舉無效票之空間分析：1992至2009

Abstract

國內外有關選舉無效票（即廢票）之研究極為稀少，主要原因是廢票率通常都不高，對選舉結果之影響不大；且無效票之產生亦被視為是一種「疏失」，而非選民「具體意志」之表達，是以很少有人探討其社會意義。 作者從1992至2009年，共考察我國44個不同類型之選舉，結果發現廢票率之空間分布有其一定之模式，並非空間隨機，而不同類型之選舉，其廢票率高低亦有差異，代表廢票率其實是可以被估計的，且其背後或許有隱涵之重要社會或政治意義。 經由廢票率之追蹤整合模型分析後，發現選舉受重視程度、制度、經社背景變數（教育年數、老年人口比及人口密度）及選舉年度，約可解釋34%之廢票率變異。此外，作者亦進一步以空間迴歸模型分析，發現廢票率可能存在正向之空間自相關（鄰近效應），或者是有遺漏具地區特性之自變數。

In the traditional election studies, the issue of non-valid votes has seldom been accepted as an important topic, because the amount of non-valid votes is usually not large enough to influence the results of elections. Besides, non-valid votes are always being considered as something created unintentionally. In this paper, the author observed 44 different elections in Taiwan, from 1992 to 2009, and found that the existing of non-valid votes is never distributed randomly. In other words, the rate of non-valid votes could be estimated, and it may have some social or political meanings deserved our attentions. The author found that the importance of an election, electoral systems, some socioeconomic variables and year of the election can explain 34% of variations of the rate of non-valid votes in a Panel Pooled OLS model. After using spatial regression models, the author discovered that the rate of non-valid votes has positive spatial autocorrelation, or maybe there are still some unobserved variables.